plotting & formatting seaborn chart from pandas da

问题:

I have a pandas dataframe al_df that contains the population of Alabama from a recent US census. I created a cumulative function that I plot using seaborn, resulting in this chart:

The code that relates to the plotting is this:

figure(num=None, figsize=(20, 10))

plt.title('Cumulative Distribution Function for ALABAMA population')
plt.xlabel('City')
plt.ylabel('Percentage')
#sns.set_style("whitegrid", {"ytick.major.size": "0.1",})
plt.plot(al_df.pop_cum_perc)

My questions are: 1) How can I change the ticks, so the yaxis shows a grid line every 0.1 units instead of the default 0.2 shown? 2) How can I change the x axis to show the actual names of the city, plotted vertically, instead of the "rank" of the city (from the Pandas index)? (there are over 300 names, so they are not going to fit well horizontally).

回答1:

For question 1) ,add:

plt.yticks(np.arange(0,1+0.1,0.1))

Question 2), I found this in the matplotlib gallery: ticks_and_spines example code

回答2:

The matplotlib way would be to use MutlipLocator. The second one is also straight forward

from matplotlib.ticker import *
plt.plot(range(10))
ax=plt.gca()
ax.yaxis.set_major_locator(MultipleLocator(0.5))
plt.xticks(range(10), list('ABCDEFGHIJ'), rotation=90) #would be range(3xx), List_of_city_names, rotation=90
plt.savefig('temp.png')

回答3:

After some research, and not been able to find a "native" Seaborn solution, I came up with the code below, partially based on @Pablo Reyes and @CT Zhu suggestions, and using matplotlib functions:

from matplotlib.ticker import *
figure(num=None, figsize=(20, 10))

plt.title('Cumulative Distribution Function for ALABAMA population')
plt.xlabel('City')
plt.ylabel('Percentage')
plt.plot(al_df.pop_cum_perc)

#set the tick size of y axis
ax = plt.gca()
ax.yaxis.set_major_locator(MultipleLocator(0.1))

#set the labels of y axis and text orientation
ax.xaxis.set_major_locator(MultipleLocator(10))
ax.set_xticklabels(labels, rotation =90)

The solution introduced a new element "labels" which I had to specify before the plot, as a new Python list created from my Pandas dataframe:

labels = al_df.NAME.values[:]

Producing the following chart:

This requires some tweaking, since specifying a display of every city in the pandas data frame, like this:

ax.xaxis.set_major_locator(MultipleLocator(1))

Produces a chart impossible to read (displaying only x axis):